Please also see Egg Study Redux: Correcting the Stats
Dr. J. David Spence of Ontario’s Stroke Prevention and Atherosclerosis Research Center and two of his colleagues have just published a study in the journal Atherosclerosis purportedly showing that eating egg yolks is almost as bad for our arteries as smoking cigarettes.
This paper is more like an aggregation of clinical notes than what we would expect from something we’d call a “study.” In 1995, they began measuring total plaque in the carotid arteries of patients who were referred to their vascular prevention clinics. Prior to the year 2000, they gave the patients a particular lifestyle questionnaire that they tell us very little about. In 2000, their clientele changed. After this point, patients came to them on an “urgent” basis because they had just suffered a stroke or a transient ischemic attack. They gave these patients “a more limited set of lifestyle questions,” which they also tell us very little about.
They asked these patients not only to estimate how often they eat eggs, but to remember for how many years they had been eating eggs at a similar frequency over their lifetime. This allowed them to calculate the number of “egg-yolk years” each patient had accumulated. We should find it unsurprising, then, that patients who had accumulated the most “egg-yolk years” not only ate more eggs, but were also older. Here’s a graph I made from the data in Table 2 showing the astonishing fact that people who had been eating more eggs for a longer time had also been alive for a longer time:
Dr. Spence and his colleagues offer us no validation of their questionnaire’s ability to estimate lifelong egg consumption, nor of the ability of recent victims of stroke and transient ischemic attack to accurately recall their dietary habits. It nevertheless seems likely to me that people over the age of 65 had indeed been eating eggs for longer than people under the age of 55, so the questionnaire is unlikely to be entirely without merit.
It may seem obvious that we have a problem here. Arterial plaque accumulates with age. Couldn’t an association between “egg-yolk years” and arterial plaque simply reflect this accumulation with age? The authors solved this little problem by making a statistical adjustment for age. After correcting for age, a greater accumulation of egg-yolk years was associated with a greater amount of plaque.
This raises the question of why they didn’t just take the simpler approach of comparing those who reported eating eggs more frequently to those who reported eating eggs less frequently. Surely the patients could more accurately recall their recent intake of eggs than their lifetime consumption of eggs, particularly if they had simply been asked something like, “how long have you been eating eggs?” The reason for using “egg-yolk years” adjusted for age instead of “eggs per week” can be found in this statement:
The total plaque area among people who consumed 2 or fewer eggs per week (n = 388) was 125 +/- 129.62 mm2, whereas it was 132.26 +/- 142.48 mm2 in those consuming 3 or more eggs per week (n = 603).
If we put this into graphic form and do our own statistical test, we can see that the difference is clearly small and not statistically significant:
The most disappointing thing about this collection of clinical notes, however, is that time itself is never considered a confounder. Are we really supposed to believe that the switch in the year 2000 to “urgent” clientele who had just suffered from transient ischemic attack or stroke makes no difference? Is it not possible that these “urgent” cases tended to have more plaque? And is it not possible that egg consumption was higher after the year 2000 than before? Here’s a graph of changes in Canadian egg consumption over time from this paper, with a few markings I have added:
The changes in this graph are exaggerated because the vertical axis hits a bottom at fourteen rather than zero. Nevertheless, it would appear that Dr. Spence and his colleagues began collecting their clinical notes in 1995 when Canadian egg consumption, estimated from food disappearance data, had reached a bottom. After this point, the egg’s reputation began recovering and consumption increased throughout the time period in which these clinicians collected their data. The “urgent” cases populating this data after 2000 may well have been eating a few more eggs than the less urgent cases that populated the data from 1995-2000. They may have been less likely to recall their diet accurately having just recently suffered from a transient ischemic attack or stroke, and they may have had a tendency to exaggerate the magnitude or duration of their egg consumption if they thought eggs may have been to blame for their medical condition.
I’m going to continue eating eggs, but if this study has made an egg-free convert out of any of you, please let me know why in the comments!